Send approval requests to multiple people simultaneously and proceed when a minimum number approve. Faster than serial chains for peer-level decisions.
Rendering diagram…
Use parallel approval when:
All approvers receive the request at the same time. The workflow tracks responses and proceeds once the threshold is met.
| Configuration | Behavior |
|---|---|
| All must approve | Wait for every approver. Any rejection = rejected. |
| Majority rules | Wait for >50% to approve. If majority reject, rejected. |
| First response wins | The first approver's decision is final. Others are cancelled. |
| N of M | Require exactly N approvals out of M approvers (e.g., 2 of 5). |
Determine who participates. Sources:
Decide the completion criteria:
threshold = 2 // minimum approvals needed
total_approvers = 3 // total people asked
timeout_days = 5 // max wait time
Create a task for each approver simultaneously. Each task should include:
As responses come in, evaluate the threshold:
| Scenario (2-of-3 threshold) | Approved | Rejected | Pending | Result |
|---|---|---|---|---|
| First response: Approve | 1 | 0 | 2 | Continue waiting |
| Second response: Approve | 2 | 0 | 1 | Approved — cancel pending tasks |
| Second response: Reject | 1 | 1 | 1 | Continue waiting |
| All responded | 1 | 2 | 0 | Rejected |
When the threshold is met:
If approvers can see each other's votes, you introduce bias. Consider hiding intermediate results until the threshold is met, especially for sensitive decisions.
Route purchase requests, expense reports, and financial approvals through different approval chains based on the dollar amount. Low-value requests skip senior approvers; high-value requests get more scrutiny.